I have a solution running on an R3.ExtraLarge Amazon instance with 50 users all of whom are using webclient. Periodically we are seeing a rise is CPU usage, and I am guessing this is also Memory usage although Amazon does not report memory metrics in an easily accessible way. Does anyone have any tips on how/where to start tracking down memory leaks of this nature ?
I have tried:
- Performance Data - this is not showing anything abnormal
- Database is static over the period so we are not getting any spikes in traffic
- I can see nothing in the logs around the time the memory starts to grow
- Growth is over 2-3 hours from 5-8% to 90+% CPU IF the machine/servoy is rebooted then CPU usage comes back to normal with same user base and stays there for several hours
- There is nothing in the usage pattern to suggest a specific event causes the problem
- The CPU is stable over night and drops down to 1-2% even if the users remain logged in, however IF rebooted before users start its remain stable to mid afternoon if not its maxed out by 10:30/11
TBH I am now lost for ideas where to look to solve this and am struggling to see how I can expand this solution using Servoy if its unable to handle more than 50 odd users. We need it to double over the next few months at least and the client is justifiably worried that Servoy may not be the right product for this next stage. The application is complex and has 300+ forms so a change from Servoy is far from ideal for anyone.
Best
Gordon
